Chandelier cells represent a unique type of cortical γ--aminobutityric acidergic interneuron whose axon terminals (Ch-terminals) only form synapses with the axon initial segments of some pyramidal cells. Here, we have used immunocytochemistry for the high-affinity plasma membrane transporter GAT-1 and the calcium-binding protein parvalbumin to analyze the morphology and distribution of Ch-terminals in the mouse cerebral cortex and claustroamygdaloid complex. In general, 2 types of Ch-terminals were distinguished on the basis of their size and the density of the axonal boutons that made up the terminal. Simple Ch-terminals were made up of 1 or 2 rows of labeled boutons, each row consisting of only 3–5 boutons. In contrast, complex Ch-terminals were tight cylinder-like structures made up of multiple rows of boutons. Simple Ch-terminals were detected throughout the cerebral cortex and claustroamygdaloid complex, the complex type was only occasionally found in certain regions, whereas in others they were very abundant. These results indicate that there are substantial differences in the morphology and distribution of Ch-terminals between different areas and layers of the mouse cerebral cortex. Furthermore, we suggest that the distribution of complex Ch-terminals may be related to the developmental origin of the different brain regions analyzed.
In the mammalian cerebral cortex, the diversity of γ-aminobutityric acidergic (GABAergic) interneurons has been demonstrated through their morphological, electrophysiological, and molecular properties. However, the contribution of each type of interneuron to cortical microcircuits in the various cortical areas, layers, and mammalian species is at present poorly understood (e.g., reviewed in Monyer and Markram 2004; DeFelipe et al. 2005; Yañez et al. 2005; Dumitriu et al. 2007). Together with basket cells, Chandelier cells represent fast-spiking GABAergic interneurons that express the calcium-binding protein parvalbumin (PV). These cells help to control the timing of hyperpolarizations, regulating the firing of pyramidal cells and shaping the network output and the rhythms generated in different states of consciousness (Cobb et al. 1995; Klausberger et al. 2003, 2004; Whittington and Traub 2003; Howard et al. 2005; Somogyi and Klausberger 2005; Inda et al. 2006). The axon terminals of chandelier cells form short vertical rows of boutons (Ch-terminals) resembling candlesticks (Szentágothai and Arbib 1974; Jones 1975), and they specifically contact the axon initial segment (AIS) of cortical principal cells (for a review, see DeFelipe 1999), a region critical for the generation of axon potentials (e.g., Stuart and Sakmann 1994). Thus, chandelier cells are thought to exert a strong inhibitory effect on pyramidal cell output (reviewed in Miles et al. 1996; DeFelipe 1999; Howard et al. 2005; but see Szabadics et al. 2006). Inhibitory inputs to the AIS of pyramidal cells from Ch-terminals have been demonstrated in different mammalian species, although they are not homogeneously distributed. Indeed, they may differ both in their density (i.e., the number of boutons that make up the Ch-terminals) and in their chemical content, as a function of age, species, cortical region and layer, or according to the projection target of the pyramidal cells (see Inda et al. 2007 and references therein). In fact, dramatic differences between distinct areas and layers of the human neocortex were identified in systematic studies using immunocytochemical techniques to analyze the density and distribution of Ch-terminals containing the high-affinity plasma membrane transporter GAT-1 (Inda et al. 2007). However, the extent to which this differential distribution may be extrapolated to other mammalian species must be explored because the relative abundance or characteristics of particular types of interneurons may vary between species (Yáñez et al. 2005; DeFelipe et al. 2005).
Current research on interneurons increasingly relies on the use of transgenic mice with genetic tags for particular populations of interneurons. Combined anatomical, physiological, and molecular analysis of these animals provides valuable data to clarify the roles played by each particular type of interneuron in cortical microcircuits (Monyer and Markram 2004; Dumitriu et al. 2007). In addition, intense research in mice over recent years has focused on the developmental processes responsible for generating interneuron diversity in the neocortex (Butt et al. 2005; Flames and Marin 2005). However, it is remarkable that no systematic studies have yet been performed on the possible regional differences in the morphology and distribution of Ch-terminals in the mouse cerebral cortex. Because alterations in chandelier cells in humans are associated with important neurological disorders, such as epilepsy (DeFelipe 1999; Wittner et al. 2001; Arellano et al. 2004) and schizophrenia (Lewis et al. 2005), such studies could be useful to interpret the specific alterations in chandelier cells that might contribute to these disorders in rodent models of these diseases (Chen et al. 2004; Cobos et al. 2005; Lewis et al. 2005; Peters et al. 2005; Nabeshima et al. 2006; Romero et al. 2007; Smart et al. 1998; Zhu et al. 2004).
In the present study, we have used GAT-1 immunocytochemistry to analyze the distribution of Ch-terminals in various areas of the archicortex (hippocampal formation), paleocortex (olfactory or piriform cortex), and neocortex, derived from distinct progenitor domains of the developing pallial telencephalon (medial, lateral, and dorsal pallium, respectively), and in the claustroamygdaloid complex which is of a mixed developmental origin (Puelles et al. 2000; Campbell 2003; Puelles and Rubenstein 2003). In addition, to further analyze the morphology of Ch-terminals and their relationship with the AIS of pyramidal cells, we analyzed double-immunostained brain sections to study the distribution of PV and phosphorylated IkBα (pIkBα), a molecule involved in the NF-kB signaling pathway that accumulates at the AIS (Schultz et al. 2006; Sanchez-Ponce et al. Forthcoming). The results indicate that there are remarkable differences in the morphology and distribution of Ch-terminals between areas and layers of the mouse pallial telencephalon.
Materials and Methods
C57BL/6 mice (n = 8 males, aged between 30 and 32 days) were sacrificed by administering a lethal intraperitoneal injection of sodium pentobarbital, and they were then perfused intracardially with saline solution followed by 4% paraformaldehyde in 0.1 M phosphate buffer (PB), pH 7.4. All experiments were performed in accordance with the guidelines established by the European Union regarding the use and care of laboratory animals. The brain of each animal was removed, postfixed by immersion in the same fixative for 24 h at 4 °C, and serial coronal sections (50 μm thick) were obtained with the aid of a Vibratome (St Louis, MO). The sections then were batch processed for immunocytochemical staining and some sections adjacent to those used for immunocytochemistry were Nissl stained.
Sections were first treated for 30 min with a solution of 0.5% hydrogen peroxide and 50% ethanol in PB to inactivate the endogenous peroxidases. Subsequently, the sections were rinsed in PB and preincubated for 1 h at room temperature in a stock solution containing 3% normal goat or horse serum (Vector Laboratories, Burlingame, CA) in PB with Triton X-100 (0.25%). Thereafter, the sections were incubated for 48 h at 4 °C in the same stock solution containing the rabbit anti-GAT-1 antiserum (1:2000; Chemicon, Temecula, CA), rabbit antiphospho (Ser32)-IKBα (1:1000; Cell Signaling Technology, Boston, MA), mouse anti-PV (1:4000, Swant, Bellinzona, Switzerland), or the mouse antineuron-specific nuclear protein (NeuN, 1:4000; Chemicon) antibody. The sections were washed in PB, incubated in biotinylated horse anti-mouse or goat anti-rabbit secondary antibodies (1:200; Vector Laboratories), and processed using the Vectastain ABC immunoperoxidase kit (Vector Laboratories). Antibody labeling for conventional light microscopy was visualized with 0.05% 3,3′diaminobenzidine tetrahydrochloride (Sigma, St Louis, MO, USA) and 0.01% hydrogen peroxide. The sections were rinsed in PB, mounted on glass slides, dehydrated, cleared with xylene, and coverslipped. Controls were included in all the immunocytochemical procedures, either by replacing the primary antibody with preimmune goat or horse serum in some sections, by omitting the secondary antibody, or by replacing the secondary antibody with an inappropriate secondary antibody. No significant immunolabeling was detected under these control conditions. Light microscopy images were captured using a digital camera (Olympus DP70) attached to a BX51 Olympus microscope. Adobe Photoshop 7.0 software was used to generate the figures (Adobe Systems Inc., San Jose, CA).
The density of complex Ch-terminal was estimated by counting them at a magnification of ×400 (using a ×40 objective) in at least ten 36 300 μm2 regions selected at random from each of the different brain areas analyzed. However, in the CA1–3 fields and the dentate gyrus, the 10 regions analyzed were 7 500 μm2. The Neurolucida package (MicroBrightField, Williston, VT) was used to plot GAT-1-immunoreactive (-ir) Ch-terminals.
Sections were double labeled by incubating them in a combination of rabbit antiphospho-IKBα (Ser32) and mouse anti-PV antibodies in PB with Triton X-100 (0.25%) and 1.5% normal goat and horse serum. The sections were then rinsed and incubated for 2 h at room temperature in biotinylated goat anti-rabbit secondary antibodies (1:200, Vector). After rinsing in PB, the sections were incubated for 2 h at room temperature in Alexa Fluor 488-conjugated streptavidin (1:2 000; Molecular Probes, Eugene, OR) and Alexa 594 conjugated horse anti-mouse (1:2 000; Molecular Probes). The sections were then washed, mounted in antifade mounting medium (Invitrogen/Molecular Probes, Eugene, OR), and examined with a Leica (Cambridge, UK) TCS 4D confocal laser scanning microscope. Z sections were recorded at 1- to 2-μm intervals through separate channels (Scanware, Leica). Subsequently, Adobe Photoshop 7.0 software (Adobe Systems Inc.) was used to construct composite images from each optical series by combining the images recorded through both channels.
The different telencephalic areas examined were identified on the basis of the anteroposterior and mediolateral coordinates of Hof et al. (2000) and according to Paxinos and Franklin (2001). The nature of these structures was later confirmed by analyzing the distinctive cytoarchitectonic features of the areas in adjacent Nissl-stained or NeuN-immunostained sections.
The observations on the distribution of Ch-terminals are presented according to the developmental origin of the cortical structures in the different histological territories of the developing mouse pallium (medial, dorsal, and lateral pallium; Figs. 1–5). The medial pallial cortical derivatives analyzed in the present study (Figs. 2 and 5) involved the 3-layered allocortical areas (archicortex or hippocampal formation) including the dentate gyrus, hippocampus proper, subiculum, and entorhinal cortex. The dorsal pallial cortical derivatives analyzed included different neocortical or isocortical areas: the primary and secondary sensory areas (visual, auditory, and somatosensory), the primary and secondary motor areas, the associative (frontal, parietal, and temporal), orbital, prelimbic, infralimbic, dorsal peduncular, cingulate, insular, and retrosplenial cortical areas (Figs. 1, 3, and 5). Finally, the lateral pallial derivatives included the piriform cortex (Figs. 1, 4, and 5). In addition, we analyzed the distribution of Ch-terminals in the claustroamygdaloid complex, but due to the mixed developmental origin of its different regions, we considered this complex separately (Figs. 4 and 5). The regions of the claustroamygdaloid complex examined include a group of cortical layered structures and nonlayered nuclei derived from the following structures: the lateral pallium (the posterolateral cortical amygdala, the dorsal part of the basolateral nucleus of the amygdala, and the dorsolateral claustrum); the ventral pallium (anterior and posteromedial cortical amygdala, posterior part of the basomedial nucleus, ventral part of the basolateral nucleus of the amygdala, lateral nucleus of the amygdala, amydalohippocampal area, ventromedial claustrum, and the endopiriform nuclei); and the subpallium (central and medial nuclei of the amygdala).
In all the areas examined, numerous terminal-like puncta were labeled in the neuropil by GAT-1 immunocytochemistry and around unstained cell bodies (Figs. 1–6). However, as described previously in the cerebral cortex of other species, the elements that are most intensely labeled are the Ch-terminals (DeFelipe and Gonzalez-Albo 1998; Conti et al. 2004; Inda et al. 2007). GAT-1-immunoreactive (-ir) Ch-terminals were found in many cortical and noncortical areas but they had very different morphologies and distributions in the different areas (see below). In general, 2 types of Ch-terminal were observed that were differentiated by their size and the density of axonal boutons that made up the Ch-terminals. For the sake of simplicity, these terminals will be referred to as “simple” and “complex” based on the earlier definitions from Golgi-stained material (e.g., Fairén and Valverde 1980; DeFelipe et al. 1985). Simple Ch-terminals were made up of 1 or 2 rows of labeled boutons, each row consisting of only 3 to 5 boutons. This kind of terminal was difficult to visualize but they appeared to be distributed throughout the cortex in double-labeling experiments (see below). By contrast, complex Ch-terminals were readily detected as tight cylinder-like structures comprised of multiple rows of boutons, and they were present in specific areas and layers. Thus, we will mainly refer to complex Ch-terminals unlike otherwise specified.
The pIκBα immunostaining was characterized by the strong labeling of numerous AISs in all layers and areas (Figs. 2–4, and 7). The somata and proximal portions of dendrites of some neurons were also immunostained but generally, less intensely than the AIS. These AISs were identified as short, thin, and smooth processes with a characteristic “eyelash-like” appearance that extended distally from the soma (arrows in Figs. 2B,D,F; 3B,D,F,H; 4B,D,F,H; and 7A,D,G,J). In all the areas studied, pIκBα-ir AIS were so numerous that although no quantitative studies have been performed, it seemed that the AIS of every neuron was immunoreactive for pIκBα. Therefore, the distribution of pIκBα-ir AIS was not homogeneous across different areas and layers, but rather, it varied with the neuronal density. In addition, differences in the size and orientation of the pIκBα-ir AISs also existed in the different areas and layers.
In the present study, we used pIκBα immunostaining to confirm and further characterize the innervation of AIS in the different areas examined. More specifically, simple Ch-terminals are often difficult to distinguish from other punctate structures and processes identified as containing GAT-1 or PV, particularly when the labeling of these elements in the neuropil is very dense or when the neuron packing is very high (e.g., the granule cell layer of the dentate gyrus, Fig. 2A,B). Therefore, we used pIκBα immunostaining (in double-labeling experiments) as a marker of the AIS in order to verify the existence of simple Ch-terminals in those areas where they were hard to detect (Fig. 7G–L). Furthermore, pIκBα immunostaining is an excellent tool to interpret the possible relationship between variations in the density and morphology (length) of Ch-terminals. For example, to examine whether cortical regions with longer pIκBα-ir AISs show more complex Ch-terminals or whether all pIκBα -ir AISs are innervated by GAT-1-ir (or PV-ir) Ch-terminals.
Distribution of GAT-1-ir Ch-Terminals in Cortical Areas Derived from the Medial Pallium: Archicortex (Hippocampal Formation)
Immunocytochemistry for GAT-1 identified numerous terminal-like punctate structures in the neuropil that surrounded unlabeled cell bodies in the various fields of the hippocampal formation, including: the dentate gyrus, hippocampus proper, subiculum, presubiculum, parasubiculum and entorhinal cortex, and the perirhinal cortex. In the dentate granule cell layer, GAT-1-ir Ch-terminals were mainly of the simple type and they were difficult to distinguish from the punctate staining (Figs. 2A and 5). As shown in adjacent sections immunostained for pIκBα (Fig. 2B), the labeled AISs were very thin and extended tortuously between the densely packed granule cell somata, making it even more difficult to identify Ch-terminals with GAT-1 or PV immunocytochemistry. In the “stratum pyramidale” of CA1–CA3 fields, both simple and complex Ch-terminals were observed as well as the intense GAT-1-ir perisomatic punctuate staining (Figs. 2C, 5, and 6G). Complex Ch-terminals were more abundant in the CA1 than in the CA3 (Fig. 5, Table 1). Less dense GAT-1-ir terminal-like puncta were detected in the neuropil, and as expected, no GAT-1-ir Ch-terminals were found in the molecular and polymorphic layers of the dentate gyrus or in the “strata oriens,” “radiatum,” and “lacunosum-moleculare” of Ammon's horn. In the pyramidal cell layer of the subiculum, presubiculum, and parasubiculum, GAT-1 immunostaining revealed the presence of simple Ch-terminals and sparsely distributed complex Ch-terminals. However, both immunostaining of perisomatic boutons and of the Ch-terminals was less intense than in the CA fields (Fig. 5). Finally, in the entorhinal cortex (Figs. 2E, 5, and 6D) and the perirhinal cortex (Figs. 1G and 5), complex GAT-1-ir Ch-terminals were very prominent, and except for layer I, they were labeled intensely in all cortical layers (Figs. 2E, 5, and 6D). In the entorhinal cortex, pIκBα-ir AISs were also very prominent (Fig. 2F and see the relative abundance of complex Ch-terminals in these regions in Table 1).
|Cerebral cortex||Claustroamygdaloid complex|
|Archicortex (medial pallium)||Neocortex (dorsal pallium)||Paleocortex (lateral pallium)||Lateral pallium||Ventral pallium||Subpallium|
|Cerebral cortex||Claustroamygdaloid complex|
|Archicortex (medial pallium)||Neocortex (dorsal pallium)||Paleocortex (lateral pallium)||Lateral pallium||Ventral pallium||Subpallium|
Note: An estimation of the density of complex Ch-terminal was made by counting them in regions of 36 300 μm2 from each different brain area analyzed. In the CA1–3 fields and dentate gyrus, these estimates were derived from regions of 7 500 μm2. The mean values are represented by the following signs: –, 0 or occasional; +, from 1 to 5; ++, from 5 to 10; and +++, more than 10. ACo, anterior cortical amygdala; AHiAL, amygdalohippocampal area anterolateral part; AHiPM, amygdalohippocampal area posteromedial part; AID, agranular insular cortex dorsal part; AIP, agranular insular cortex posterior part; AIV, agranular insular cortex , ventral part; APC, associative parietal cortex; Au1, primary auditory cortex; AuD, secondary auditory cortex , dorsal area; AuV, secondary auditory cortex , ventral area; BLD, basolateral nucleus of the amydala , dorsal part; BLV, basolateral nucleus of the amydala ventral part; BMP, basomedial nucleus of the amydala ,posterior part; CA1, field CA1 of hippocampus; CA2, field CA2 of hippocampus; CA3, field CA3 of hippocampus; cg, cingulate cortex; CE, central nucleus of the amygdala; Cld, claustrum , dorsolateral part; Clv, claustrum ,ventromedial part; DEn, dorsal endopiriform nucleus; DG, dentate gyrus; DI, dysgranular insular cortex; DP, dorsal peduncular cortex; DTT, dorsal tenia tecta; Ect, ectorhinal cortex; FrA, frontal association cortex; IL, infralimbic cortex; GI, granular insular cortex; La, lateral nucleus of the amygdala; LEnt, lateral entorhinal cortex; LO, lateral orbital cortex; M1, primary motor cortex; M2, secondary motor cortex; MO, medial orbital cortex; ME, medial nucleus of the amygdala; MEnt, medial entorhinal cortex; PaS, parasubiculum; Pir, piriform cortex; PLCo, posterolateral cortical amygdala; PMCo, posteromedial cortical amygdala; PRh, perirhinal cortex; PrL, prelimbic cortex; PrS, presubiculum; RSA, retrosplenial agranular cortex; RSG, retrosplenial granular cortex; S, subiculum; S1, primary somatosensory cortex; S2, secondary somatosensory cortex; TeA, temporal association cortex; V1, primary visual cortex; V2, secondary visual cortex; VEn, ventral endopiriform nucleus; and VO, ventral orbital cortex.
Distribution of GAT-1-ir Ch-Terminals in Cortical Areas Derived from the Dorsal Pallium: Neocortex
There were dramatic differences in the neocortex with regards to the relative abundance of complex GAT1-ir Ch-terminals (Table 1 and Fig. 5). In contrast to the abundance of terminal-like punctate structures, very few complex Ch-terminals immunostained for GAT-1 were found in the primary and secondary visual (Figs. 1A and 5), somatosensory (Figs. 1C, 3A, and 5), auditory dorsal area (Figs. 1B and 5), orbital, insular, retrosplenial (Fig. 5) motor (Figs. 3C and 5), and in the frontal (Fig. 1D) and parietal associative areas of the neocortex. Indeed, despite the prominence of AIS in the adjacent pIκBα-immunostained sections, simple GAT-1-ir Ch-terminals were mostly identified (Fig. 3B and D). In contrast, in the ventral part of the secondary auditory cortex and in the temporal associative cortex (Figs. 1E and 3E), complex GAT-1-ir Ch-terminals were observed mainly in the infragranular layers. Finally, in the cingulate (Figs. 1F and 3G), dorsal peduncular, infralimbic, and prelimbic cortical areas, complex GAT-1-ir Ch-terminals were distributed throughout layers II–VI, although they were most abundant in layer II.
Distribution of GAT-1-ir Ch-Terminals in Areas Derived from the Lateral Pallium: Paleocortex
Within the paleocortex, complex GAT-ir Ch-terminals were widely distributed in layers II and III of the piriform cortex, where they were very abundant (Table 1 and Fig. 5) and intensely immunostained (Table 1 and Figs. 1H, 4A, and 5). In contrast to the neocortex where both GAT-1-ir Ch-terminals and pIκBα-ir AISs were distributed regularly with a radial orientation, in the piriform cortex they were more randomly distributed. Despite the abundance of complex GAT-ir Ch-terminals in the piriform cortex, no complex GAT-ir Ch-terminals were found in adjacent peripaleocortical areas such as the insular cortex. Therefore, the limit between the anterior piriform cortex and the insular cortex was easily distinguished by the sharp decrease in the density of complex GAT-ir Ch-terminals.
Distribution of GAT-1-ir Ch-Terminals in the Claustroamygdaloid Complex
According to the specific patterns of gene expression (see Puelles et al. 2000; Martínez-García et al. 2002; Puelles and Rubenstein 2003; Medina et al. 2004; Moreno and Gonzalez 2006), the claustroamygdaloid complex represents the caudal part of several telencephalic histogenetic territories (lateral pallium, ventral pallium, and subpallium). These territories include both laminated (cortical) and nonlaminated (nuclear) components, some of which contained numerous complex GAT-1-ir Ch-terminals.
Derivatives of the lateral pallium.
The structures derived from the lateral pallium in the claustroamygdaloid complex include the posterolateral part of the cortical nucleus of the amygdala. In this region, large and radially oriented complex Ch-terminals were intensely labeled for GAT-1 in layers II and III (Figs 4C and 5). In addition, complex GAT-1-ir Ch-terminals were found in the nonlaminated derivatives of the lateral pallium, including the dorsolateral claustrum and the dosal part of the basolateral nucleus of the amygdala where they were more abundant and intensely stained (Table 1 and Fig. 4E).
Derivatives of the ventral pallium.
The ventral pallium gives rise to the anterior and posteromedial zones of the cortical amygdala where radially oriented complex GAT-1-ir Ch-terminals were observed in layers II and III (data not shown). In addition, nonlaminated derivatives of the ventral pallium include areas in which complex GAT-1-ir Ch-terminals were found such as the lateral, basolateral (ventral part) and basomedial (posterior part) nuclei of the amygdala, the amygdalohippocampal area, the ventromedial claustrum, and the endopiriform nuclei (Table 1).
Derivatives of the subpallium.
In contrast to the subdivisions of the claustroamygdaloid complex derived from pallial territories, no complex GAT-1-ir Ch-terminals could be identified in the subpallial amygdala, including the central amygdala (Fig. 4G) and the medial amygdala (Table 1).
There are no commercial antibodies available to perform reliable dual immunhistochemistry for GAT-1 and pIκBα. Hence, to directly analyze the relationship between the morphology of Ch-terminals and the AIS, we have used antibodies against PV in combination with pIκBα antibodies because antibodies against PV mostly recognize the same Ch-terminals as those directed against GAT-1 (see DeFelipe and Gonzalez-Albo 1998). In addition to Ch-terminals, PV immunocytochemistry stained nonpyramidal cell bodies and a dense plexus of axonal and dendritic processes in the principal cell layers of the areas examined, as shown previously (Fig. 6). Although PV and GAT-1 did not colocalize in all Ch-terminals, PV-ir Ch-terminals were generally distributed in a morphological, regional, and laminar pattern similar to the staining for GAT-1 (Fig. 6). Nevertheless, Ch-terminals were more difficult to distinguish by PV immunostaining than by GAT-1 immunostaining, mainly due to the presence of numerous PV-ir dendritic and axonal processes that masked the Ch-terminals. The scarcity of complex Ch-terminals in some areas of the paleocortex, archicortex, or neocortex, initially observed by specific GAT-1 immunostaining was confirmed by PV immunostaining. However, experiments with double labeling for pIκBα and PV (Fig. 7) revealed that the lack of complex PV-ir Ch-terminals does not imply the absence of PV innervation of the AIS. In the cortical areas where no complex Ch-terminals were observed, the use of Z-stack confocal microscopy reconstructions revealed that some of the pIKBα-ir AISs were innervated by a few or no PV-ir boutons, suggesting that the density of GABAergic synaptic contacts with the AIS formed by complex Ch-terminals was low in these cortical regions (Fig. 7G–I). Finally, in all areas where complex Ch-terminals could be identified (either by GAT-1 or PV immunostaining), complex PV-ir Ch-terminals mostly innervated the distal portion of the pIκBα-ir AIS of pyramidal cells (Fig. 7J–L). In contrast, adjacent pIκBα-ir AIS were seen to be innervated by relatively few PV-ir boutons or they lacked such innervation. Therefore, in these latter regions some AIS were innervated by complex Ch-terminals, others by simple Ch-terminals, whereas others were apparently not innervated by PV-ir chandelier cells at all.
The main finding of this study is that although chandelier cells populate the entire pallial telencephalon, there are substantial differences in the distribution of complex GAT-1-ir Ch-terminals in the distinct regions and layers of both the cerebral cortex and claustroamygdaloid complex. In the cerebral cortex, complex GAT-1-ir Ch-terminals were generally scarce in areas derived from the dorsal pallium (neocortex). By contrast, these terminals were particularly dense in the piriform and entorhinal cortex, which derive from the lateral and the medial pallium, respectively. In the claustroamygdaloid complex, GAT-1-ir complex Ch-terminals were found in those areas derived from the pallium but not in the nuclei derived from the subpallium.
Distribution of Ch-Terminals in Nonneocortical Telencephalic Areas
This study indicates that complex GAT-1-ir Ch-terminals are the most widely distributed and intensely stained in the piriform cortex, which is a paleocortical area derived from the lateral pallium. In the archicortex, complex GAT-1-ir Ch-terminals were most prominent in the entorhinal and perirhinal cortices, followed by the subicular complex and then by the hippocampal formation. In this latter structure, these terminals were found in the stratum pyramidale of the CA1–CA3 and only occasionally in the granule cell layer of the dentate gyrus. These results are in agreement with previous light and electron microscopy studies in the hippocampal formation of humans that noted the presence of chandelier cells in the hippocampus and dentate gyrus, mainly in Golgi-stained and PV-immunostained material (humans: Seress et al. 1993; Wittner et al. 2001; Arellano et al. 2004; rats: Kosaka 1983; Somogyi, Kisvárday, et al., 1983; Somogyi, Nunzi, et al., 1983; Somogyi et al. 1985; Soriano and Frotscher 1989; Ribak et al. 1990; Soriano et al. 1990; Li et al. 1992; Gulyás et al. 1993; Halasy and Somogyi 1993; Buhl, Halasy, Somogyi 1994; Buhl, Han, et al. 1994;,Han et al. 1993; reviewed in Freund and Buzsaki 1996)
In addition, here we demonstrated the presence of numerous complex GAT-1-ir Ch-terminals in certain regions of the claustroamygdaloid complex, including the pallial part of the amygdala. This area contains excitatory (presumably glutamatergic) projection neurons, like the cerebral cortex and the claustrum (Swanson and Petrovich 1998; Swanson 2000), and it includes structures that originate from the lateral and ventral pallium (McDonald 1998; Puelles et al. 2000, 2001; Martínez-García et al. 2002; Sah et al. 2003; Stenman et al. 2003; Medina et al. 2004; Remedios et al. 2004; Tole et al. 2005; García-López et al. 2008). This observation is in line with previous studies in the rat showing dense PV-ir Ch-terminals in these regions (Kemppainen and Pitkanen 2000). In contrast, complex GAT-1-ir Ch-terminals were not observed in the regions of the amygdala derived from the subpallium (centromedial amygdaloid area), which are similar to the striatum and pallidum and that are rich in GABAergic projection neurons (Swanson and Petrovich 1998; Swanson 2000,Puelles et al. 2000, 2001; Sah et al. 2003; Stenman et al. 2003). Thus, the presence of complex Ch-terminals in the claustroamygdaloid complex might be related to the presence or absence of glutamatergic projecting neurons.
Morphology and Distribution of Ch-Terminals in the Neocortex
Simple and complex Ch-terminals
Previous studies have identified both simple and complex Ch-terminals in the visual cortex of the rat (Peters et al 1982) and the cat (Fairén and Valverde 1980), the somatosensory cortex of the monkey (DeFelipe et al. 1985), and in various areas of the human neocortex (Inda et al. 2007). However, the finding that complex Ch-terminals are essentially restricted to certain areas and layers in the mouse neocortex probably reflects the existence of species-specific differences in the distribution of simple and complex Ch-terminals. Simple Ch-terminals are certainly harder to identify in PV- or GAT-1-immunostained sections because it is difficult to distinguish them within the dense plexus of immunostained elements. Nevertheless, double-staining tissue for PV and pIκBα makes it relatively easy to define labeled terminals that innervate the AIS (see Fig. 7). Still, it is possible that not all axoaxonic terminals originate from chandelier cells because other types of interneurons occasionally establish synapses with the AIS of pyramidal neurons (Fariñas and DeFelipe 1991; Gonchar et al. 2002). Thus, what we consider simple Ch-terminals may include some terminals from other types of interneurons.
Density of Ch-Terminals in Different Areas
No previous studies have examined the distribution of Ch-terminals in the various cortical areas of the mouse or rat. However, in the monkey (Lewis et al. 1989; Akil and Lewis 1992; Conde et al. 1996; Elston and Gonzalez-Albo 2003) and in humans (Inda et al. 2007), it has been shown that there is generally a higher density of terminals innervating the AIS in association areas of the neocortex than in primary sensory areas. Indeed, quantitative studies of the density of GAT-1-ir Ch-terminals in the human neocortex have shown remarkable differences both between areas and also between distinct layers (Inda et al. 2007). In the primary sensory cortex, including the visual cortex, complex GAT-ir (or PV-ir) Ch-terminals are virtually absent in the mouse although they are present in the human primary sensory cortex, albeit in a relatively small numbers (Inda et al. 2007). In addition, in the motor areas and frontal and parietal associative cortex, very few complex Ch-terminals were found in mice, whereas this kind of terminals were frequently found in these areas of the human neocortex (Inda et al. 2007). In the mouse secondary auditory and temporal associative areas, complex GAT-1-ir Ch-terminals were virtually restricted to the infragranular layers. Furthermore, chandelier cells express different substances depending on the cortical region and layer (e.g., del Río and DeFelipe 1997; Arellano et al. 2002; for a review, see DeFelipe 1999). In this regard, the development of cortical interneurons occurs in a region- and cell subtype-specific manner (Xu et al. 2004; Butt et al. 2005; Liodis et al. 2007). Thus, it is possible that there are specific mechanisms that guide the migration of different morphological and neurochemical types of chandelier cells, restricting or concentrating them in particular regions of the cerebral cortex.
Possible Functional Significance
The substantial differences found between areas and layers are probably related to their different functional attributes, and they suggest that chandelier cells contribute differentially to cortical circuits according to their location. For example, in the cat visual cortex the AISs of pyramidal cells projecting to the thalamus established extremely few synapses (from 1 to 5) when compared with callosal pyramidal cells (from 16 to 23) and ipsilateral corticocortical pyramidal cells (from 22 to 28: Fariñas and DeFelipe 1991). Because in the mouse neocortex complex GAT-1-ir Ch-terminals were observed in large numbers in the infragranular layers of the auditory and temporal associative areas alone, it is possible that this type of Ch-terminal is related to the innervation of a subpopulation of pyramidal cells that project to particular sites.
Pyramidal neurons also show remarkable differences regarding the complexity of the dendritic trees and number of dendritic spines in each area (i.e., the number presumptive excitatory glutamatergic inputs; for a review, see Elston 2002, 2003). However, there does not appear to be a general rule regarding the relationships between pyramidal morphology and the density of AIS innervation. For example, complex Ch-terminals are relatively few in area 18 of the human cortex, whereas in area 10 of the prefrontal cortex they are very abundant (Inda et al. 2007). In these cortical areas, the number of dendritic spines of the presumptive pyramidal neurons innervated by Ch-terminals is relatively low in area 18 when compared with area 10 (Elston et al. 2001). Thus, pyramidal cells in the visual cortex theoretically integrate fewer synaptic inputs than in the prefrontal cortex and, due to a lower density of GABAergic Ch-terminals, they are subjected to a less powerful inhibitory control at the level of the AIS. However, in the mouse, Ch-terminals are mostly simple in all the neocortical areas but there are significant differences in the number of dendritic spines on pyramidal cells in the different neocortical areas (Ballesteros-Yáñez et al. 2006; Ballesteros-Yáñez, I, Bourgeois, JP, Changeux, JP, and DeFelipe, J, unpublished results). Thus, it seems that there is no relationship between the weight of the excitatory inputs to pyramidal cells and the density of the GABAergic innervation at the level of AIS, at least in the mouse neocortex.
The differences in the GABAergic inhibition of the AIS by chandelier cells also occur postsynaptically in the rat cerebral cortex. Indeed, postsynaptic AIS GABAA receptors of supragranular pyramidal cells are enriched in the α2 subunit, whereas the α3 subunit predominates in the infragranular layers (Fritschy et al. 1998). Therefore, chandelier cells seem to exert their activity through different numbers of axoaxonic terminals that might result from different degrees of chandelier cell convergence or due to chandelier cells forming simple or complex terminals (further discussed in DeFelipe et al. 1985), and through different types of postsynaptic GABAA receptors, depending on the cortical area and layer in which they are found. Further studies will be necessary to elucidate the functional significance of these microanatomical cortical differences.
Finally, our limited knowledge of the comparative morphology of particular types of interneurons in the cerebral cortex is mostly based on studies performed with the Golgi method (e.g., Fairén et al. 1984). However, the inconsistencies and incomplete staining obtained with the Golgi method makes it difficult to interpret the similarities or differences between these studies. It was evident here that complex Ch-terminals are relatively abundant in the mouse paleocortex and archicortex, structures derived from the medial and lateral pallium, respectively. Moreover, such terminals are largely absent in many neocortical areas derived from the dorsal pallium. By contrast, in studies performed in humans, numerous complex Ch-terminals were found throughout the whole cerebral cortex (see Arellano et al. 2002, 2004; Inda et al. 2007; and references therein), suggesting that evolutionary there is a trend toward an increase in the complexity of the axonal arborization of chandelier cells as the neocortex expanded and differentiated.
Ministerio de Educación y Ciencia (BFU2006-03855) to A.M. and BFU2006-13395 to J.D.
Conflict of Interest: None declared.